1 / 33

David Notkin  Autumn 2009  CSE303 Lecture 25

David Notkin  Autumn 2009  CSE303 Lecture 25. The plan. HW7 is out; new PM due date Finish last lecture. References. type& name = variable; reference: A variable that is a direct alias for another variable. any changes made to the reference will affect the original

george
Download Presentation

David Notkin  Autumn 2009  CSE303 Lecture 25

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. David Notkin Autumn 2009  CSE303 Lecture 25

  2. The plan • HW7 is out; new PM due date • Finish last lecture

  3. References • type& name = variable; • reference: A variable that is a direct alias for another variable. • any changes made to the reference will affect the original • like pointers, but more constrained and simpler syntax • an effort to "fix" many problems with C's implementation of pointers • Example: int x = 3; int& r = x; // now use r just like any int r++; // r == 4, x == 4 • value on right side of = must be a variable, not an expression/cast

  4. References vs. pointers • don't use * and & to reference / dereference (just & at assignment) • cannot refer directly to a reference; just refers to what it refers to • a reference must be initialized at declaration • int& r; // error • a reference cannot be reassigned to refer to something else int x = 3, y = 5; int& r = x; r = y; // sets x == 5, r == 5 • a reference cannot be null, and can only be "invalid" if it refers to an object/memory that has gone out of scope or was freed

  5. Reference parameters returntype name(type& name, ...) { ... } • client passes parameter using normal syntax • if function changes parameter's value, client variable will change • you almost never want to return a reference • except in certain cases in OOP

  6. const and references • const: Constant, cannot be changed. • used much, much more in C++ than in C • can have many meanings (const pointer to a const int?) void printSquare(const int& i){ i = i * i; // error cout << i << endl; } int main() { inti = 5; printSquare(i); }

  7. Strings • #include <string> • C++ actually has a class for strings • much like Java strings, but mutable (can be changed) • not the same as a "literal" or a char*, but can be implicitly converted string str1 = "Hello"; // impl. conv. • Concatenating and operators string str3 = str1 + str2; if (str1 == str2) { // compares characters if (str1 < str3) { // compares by ABC order char c = str3[0]; // first character

  8. String methods string s = "Goodbye world!"; s.insert(7, " cruel"); // "Goodbye cruel world!"

  9. String concatenation • a string can do + concatenation with a string or char*,but not with an int or other type: string s1 = "hello"; string s2 = "there"; s1 = s1 + " " + s2; // ok s1 = s1 + 42; // error • to build a string out of many values, use a stringstream • works like an ostream (cout) but outputs data into a string • call .str() on stringstream once done to extract it as a string #include <sstream> stringstream stream; stream << s1 << " " << s2 << 42; s1 = stream.str(); // ok

  10. Libraries #include <cmath>

  11. Arrays • stack-allocated (same as C): type name[size]; • heap-allocated: type* name = new type[size]; • C++ uses new and delete keywords to allocate/free memory • arrays are still very dumb (don't know size, etc.) int* nums = new int[10]; for (inti = 0; i < 10; i++) { nums[i] = i * i; } ... delete[] nums;

  12. malloc vs. new

  13. Exceptions • exception: An error represented as an object or variable. • C handles errors by returning error codes • C++ can also represent errors as exceptions that are thrown / caught • throwing an exception with throw: double sqrt(double n) { if (n < 0) { throw n; // kaboom } ... • can throw anything (a string, int, etc.) • can make an exception class if you want to throw lots of info:#include <exception>

  14. More about exceptions • catching an exception with try/catch: try { double root = sqrt(x); } catch (double d) { cout << d << " can't be squirted!" << endl; } • throw keyword indicates what exception(s) a method may throw • void f() throw(); // none • void f() throw(int); // may throw ints • predefined exceptions (from std::exception): bad_alloc, bad_cast, ios_base::failure, ...

  15. C++ classes • class declaration syntax (in .h file): class name { private: members; public: members; }; • class member definition syntax (in .cpp file): returntypeclassname::methodname(parameters) { statements; } • unlike in Java, any .cpp or .h file can declare or define any class (although the convention is still to put the Foo class in Foo.h/cpp)

  16. A class's .h file #ifndef _POINT_H #define _POINT_H class Point { private: int x; int y; // fields public: Point(int x, int y); // constructor intgetX(); // methods intgetY(); double distance(Point& p); void setLocation(int x, int y); }; #endif

  17. A class's .cpp file #include "Point.h" // this is Point.cpp Point::Point(int x, int y) { // constructor this->x = x; this->y = y; } int Point::getX() { return x; } int Point::getY() { return y; } void Point::setLocation(int x, int y) { this->x = x; this->y = y; }

  18. Simple example • A Point constructor with no x or y parameter; if no x or y value is passed, the point is constructed at (0, 0). • A translate method that shifts the position of a point by a given dx and dy. // Point.h public: Point(int x = 0, int y = 0); // Point.cpp void Point::translate(intdx, intdy) { setLocation(x + dx, y + dy); }

  19. More about constructors • initialization list: alternate syntax for storing parameters to fields • supposedly slightly faster for the compiler class::class(params) : field(param), ..., field(param) { statements; } Point::Point(int x, int y) : x(x), y(y) {} • if you don't write a constructor, you get a default () constructor • initializes all members to 0-equivalents (0.0, null, false, etc.)

  20. Multiple constructors • if your class has multiple constructors: • it doesn't work to have one constructor call another • but you can create a common init function and have both call it

  21. Constructing objects • client code creating stack-allocated object: type name(parameters); Point p1(4, -2); • creating heap allocated (pointer to) object: type* name = new type(parameters); Point* p2 = new Point(5, 17); • in Java, all objects are allocated on the heap • in Java, all variables of object types are references (pointers)

  22. A client program #include <iostream> #include "Point.h" using namespace std; int main() { Point p1(1, 2); Point p2(4, 6); cout << "p1 is: (" << p1.getX() << ", " << p1.getY() << ")" << endl; // p1 is: (1, 2) cout << "p2 is: (" << p2.getX() << ", " << p2.getY() << ")" << endl; // p2 is: (4, 6) cout << "dist : " << p1.distance(p2) << endl; return 0; // dist : 5 }

  23. Client with pointers #include <iostream> #include "Point.h" using namespace std; int main() { Point* p1 = new Point(1, 2); Point* p2 = new Point(4, 6); cout << "p1 is: (" << p1->getX() << ", " << p1->getY() << ")" << endl; // p1 is: (1, 2) cout << "p2 is: (" << p2->getX() << ", " << p2->getY() << ")" << endl; // p2 is: (4, 6) cout << "dist : " << p1->distance(*p2) << endl; delete p1; // dist : 5 delete p2; // free return 0; }

  24. Stack vs. heap objects • which is better, stack or pointers? • if it needs to live beyond function call (e.g. returning), use a pointer • if allocating a whole bunch of objects, use pointers • "primitive semantics" can be used on objects • stack objects behave use primitive value semantics (like ints) • new and delete replace malloc and free • new does all of the following: • allocates memory for a new object • calls the class's constructor, using the new object as this • returns a pointer to the new object • must call delete on any object you create with new, else it leaks

  25. Why doesn't this code change p1? int main() { Point p1(1, 2); cout << p1.getX() << "," << p1.getY() << endl; example(p1); cout << p1.getX() << "," << p1.getY() << endl; return 0; } void example(Point p) { p.setLocation(40, 75); cout << "ex:" << p.getX() << "," << p.getY() << endl; } // 1,2 // ex:40,75 // 1,2

  26. Object copying • a stack-allocated object is copied whenever you: • pass it as a parameter foo(p1); • return it return p; • assign one object to another p1 = p2; • the above rules do not apply to pointers • object's state is still (shallowly) copied if you dereference/assign *ptr1 = *ptr2; • You can control how objects are copiedby redefining the = operator for your class (ugh)

  27. Objects as parameters • We generally don't pass objects as parameters like this: double Point::distance(Point p) { intdx = x - p.getX(); intdy = y - p.getY(); return sqrt(dx * dx + dy * dy); } • on every call, the entire parameter object p will be copied • this is slow and wastes time/memory • it also would prevent us from writing a method that modifies p

  28. References to objects • Instead, we pass a reference or pointer to the object: double Point::distance(Point& p) { intdx = x - p.getX(); intdy = y - p.getY(); return sqrt(dx * dx + dy * dy); } • now the parameter object p will be shared, not copied • are there any potential problems with this code?

  29. const object references • If the method will not modify its parameter, make it const double Point::distance(const Point& p) { intdx = x - p.getX(); intdy = y - p.getY(); return sqrt(dx * dx + dy * dy); } • the distance method is promising not to modify p • if it does, a compiler error occurs • clients can pass Points via references without fear that their state will be changed

  30. const methods • If the method will not modify the object itself, make the method const: double Point::distance(const Point& p) const { intdx = x - p.getX(); intdy = y - p.getY(); return sqrt(dx * dx + dy * dy); } • a const after the parameter list signifies that the method will not modify the object upon which it is called (this) • helps clients know which methods aren't mutatorsand helps the compiler optimize method calls • a const reference only allows const methods to be called on it

  31. const and pointers • const Point* p • p points to a Point that is const; cannot modify that Point's state • can reassign p to point to a different Point (as long as it is const) • Point* const p • p is a constant pointer; cannot reassign p to point to a different object • can change the Point object's state by calling methods on it • const Point* const p • p points to a Point that is const; cannot modify that Point's state • p is a constant pointer; cannot reassign p to point to a different object • (This is not one of the more beloved features of C++.)

  32. Pointer, reference, etc.? • How do you decide whether to pass a pointer, reference, or object? Some principles: • Minimize the use of object pointers as parameters.(C++ introduced references for a reason. They are safer and saner.) • Minimize passing objects by value, because it is slow, it has to copy the entire object and put it onto the stack, etc. • In other words, pass objects as references as much as possible; but if you really want a copy, pass it as a normal object. • Objects as fields are usually pointers (why not references?). • If you are not going to modify an object, declare it as const. • If your method returns a pointer/object field that you don't want the client to modify, declare its return type as const.

  33. Questions?

More Related